Conventionally a technique that specifies a character area and a non-character area in an image and applies edge extraction processing to data in the character area while applying smoothing processing to data in the non-character area has been known.